In Situ Raman Study of Phase Stability of !‑Li3V2(PO4)3 upon Thermaland Laser HeatingNellymar Membreno,† Penghao Xiao,† Kyu-Sung Park,‡,∥ John B. Goodenough,‡,§,∥

Graeme Henkelman,*,† and Keith J. Stevenson*,†,§,∥

†Department of Chemistry and Biochemistry (105 E. 24th St. Stop A5300), ‡Materials Science and Engineering Program, §Center forElectrochemistry, and ∥Texas Materials Institute, The University of Texas at Austin, Austin, Texas 78712, United States

*S Supporting Information

ABSTRACT: Monoclinic α-Li3V2(PO4)3 has a complex 3-D metal phosphate frameworkthat provides mobility for all three lithium ions, giving it the highest gravimetric capacity(197 mAh/g) of all the transition-metal phosphates. Along with its high gravimetriccapacity, its thermal and electrochemical stability make it of great interest as a cathodematerial for lithium-ion energy storage devices. Raman spectroscopy has proven to be aunique analytical tool for studying electrode materials of lithium-ion batteries due to itsability to probe structural changes at the level of chemical bonds. In this work, thecalculated Raman spectrum of α-Li3V2(PO4)3 provided by density functional theory ispresented along with symmetry assignments for all of the calculated and observed modesthrough Raman microscopy. Furthermore, the phase stability of microcrystalline α-Li3V2(PO4)3 was studied as a function of irradiation power density. Follow-up thermalstudies confirm that two structural phase transitions, β and γ, occur at elevatedtemperatures or high irradiation power density before degradation to α-LiVOPO4 underan oxygen-rich atmosphere. Calculated and experimentally determined Raman modes for α-Li3V2(PO4)3 are in good agreement.It is also noted that careful consideration of the irradiation power density employed must be taken into account to preventmisinterpretation of Raman spectral features.

1. INTRODUCTIONTransition-metal phosphates that include LiFePO4, LiMnPO4,and Li3V2(PO4)3 have resurged as alternatives to thecommercially dominating LiCoO2 cathode owing to theirsafety, thermal stability, low cost, and competitive energydensities.1,2 The inherent stability of these materials originatesfrom the P−O bond covalency in the metal phosphateframework that houses mobile Li ions.1 Li3V2(PO4)3 can befound in either the rhombohedral (NASICON) or themonoclinic (α) lattice system at room temperature, dependingon the synthetic route, with the latter phase being the morethermodynamically stable.2

In the monoclinic lattice, the V2(PO4)3 units are alternatelypositioned perpendicular to one another, creating a slightlymore compressed structure as compared to the rhombohedralphase.3 More specifically, two distinct vanadium sites arepresent in slightly distorted VO6 octahedra that share oxygenvertices with PO4 tetrahedra (the three phosphate tetrahedracan be distinguished by their varying average P−O bondlengths).2 Within the interstitial voids of the lattice are threeunique lithium sites. Following the assignment used by Yin etal., Li(1) is coordinated to four oxygens, forming a truetetrahedron, while Li(2) and Li(3) occupy a highly distortedtetrahedron with a fifth Li−O bond.2 Along the a axis of thecrystal, Li polyhedra are linked through common oxygen,allowing for diffusion along the other directions.3 However, Leeet al. calculated lower migration energies along the [001]

direction, resulting in a highly anisotropic Li+ mobility despitenot being one-dimensional.4 Nevertheless, the ability toelectrochemically extract all three lithium ions from α-Li3V2(PO4)3 gives it the high specific gravimetric capacity of197 mAh/g. This value is the highest of all the transition-metalphosphates, making it a prospective cathode material forlithium-ion batteries and meriting a fundamental understandingof its vibrational modes through Raman microscopy.Raman microscopy is highly suitable for probing the

chemical bond structure in electrode materials used inlithium-ion energy storage devices. The microscope offers theability to focus the photon beam onto a small area (∼1 μmdiameter) of importance for electrode materials that arecomposite mixtures of conductive carbon (to enhanceelectronic conductivity) and binder.5 Raman microscopyfrequently serves as a complementary technique to X-raydiffraction (XRD)6 and powder neutron diffraction (PND) datafor crystalline materials. However, unlike diffraction methodsthat solely identify materials with long-range order, Ramanspectroscopy allows for the investigation of amorphousmaterials.5 On a more technical level, there is no need for aparticular sample preparation and the excitation source ischosen to be nondestructive. Consequently, little to no

Received: April 2, 2013Revised: May 11, 2013Published: May 14, 2013

Article

pubs.acs.org/JPCC

© 2013 American Chemical Society 11994 dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−12002

modifications are made to the sample, permitting it to befurther analyzed by other techniques.5

From an analytical perspective, Raman microscopy providesthe high sensitivity necessary to detect changes in crystalsymmetry, oxidation states, local phase inhomogeneities, andstructural order/disorder.5 This sensitivity is especially usefulfor ex situ analysis of cycled electrodes.5,7 Moreover, in situmeasurements can be easily devised by using transparentmaterials (such as glass, quartz, and sapphire) as opticalwindows for electrochemical cells as they present too weak aRaman signal to cause interference.8−10 These spectroelec-trochemical devices have served as invaluable tools forunderstanding lithium intercalation mechanisms in bothcathode and anode materials as well as for the understandingof the chemical and/or electrochemical processes that areresponsible for cell failure.5

Although structural characterization of transition-metalphosphates has predominantly been executed via Li7 magic-angle spinning nuclear magnetic resonance (MAS NMR),11,12

Mossbauer spectroscopy, and PND,13 a number of opticalspectroscopy studies have been reported for these materi-als.14−17 The commercial success of the cheap and toxicallybenign LiFePO4 has made it the most thoroughly studied of thephosphates.1 Burba et al.14 observed Raman spectral changes ofchemically delithiated LiFePO4 (0 ≤ x ≤ 1). With partialdelithiation (Li0.74FePO4), new spectral peaks were found in thev1 and v3 region of the spectrum that grew in intensity untilcomplete delithiation (FePO4). Despite not having completeassignments of the vibrations, these observations providedindirect evidence of the two-phase mechanism for theelectrochemical delithiation of LiFePO4. Paraguassu et al.18

later assigned all 36 Raman wavenumbers for LiFePO4 using acomputational simulation based on Wilson’s FG matrixmethod. The method was applied to LiFePO4 and LiNiPO4.It showed good agreement between experimental andcalculated wavenumbers (about 5% and 9% deviation,respectively).Li3M2(PO4)3 (M = Sc, Fe) compounds, isostructural to

Li3V2(PO4)3, were investigated by Kravchenko et al.17 by means

of infrared (IR) and Raman microscopy in the 77−670 Ktemperature range to understand the structural changesassociated with the superionic phase transition.7 Completespectral assignments were not made in these studies, butidentification of the sublattice lithium vibrations, found at 505cm−1 in the Raman spectrum of Li3Sc2(PO4)3, was made bycomparison of Li6 isotope enriched Li3Sc2(PO4)3 with anonenriched sample. Burba et al.7 later looked at vibrationsfor the electrochemically delithiated species (0 ≤ x ≤ 3) of bothNASICON and α-Li3V2(PO4)3 with emphasis on the (PO4)

3−

vibrations that are highly sensitive to Li+ extraction and theoxidation state of vanadium. Ex situ mid-IR spectra were foundto be in accordance with the already established complex two-phase transition mechanism proposed by Yin et al.2,12

Furthermore, the work showed that, after the first electro-chemical cycle, full reversibility of the α-Li3V2(PO4)3 structureis regained with some (PO4)

3− band broadening indicative ofsome local disordering. The investigators intended to similarlyanalyze the (PO4)

3− Raman active vibrations for α-Li3V2(PO4)3at various electrochemically delithiated stages, but the weakbands observed did not allow for further analysis. The usuallyintense phosphate peaks were thought to be broadened andweakened because of an amorphous surface layer on theparticles.

Here, the complete Raman vibrational spectrum of α-Li3V2(PO4)3 is reported and compared to the theoreticalRaman frequencies (wavenumbers) calculated with densityfunctional theory (DFT) using the Vienna ab inito simulationpackage (VASP). In addition, thermal effects as a function ofirradiation power density were studied to understand the phasestability of α-Li3V2(PO4)3 under confocal Raman measure-ments. Our findings show that, with increasing irradiationpower, α-Li3V2(PO4) transforms into the high-temperature βand γ-phases before becoming completely oxidized to LiVO-PO4. The oxidation process was further confirmed by thermalstudies with both Raman interfaced with a hot-stage anddifferential scanning calorimetry (DSC).

2. EXPERIMENTAL AND THEORETICAL METHODS!-Li3V2(PO4)3 Solid State Synthesis. For the preparation

of α-Li3V2(PO4)3, stoichiometric amounts of Li2CO3 (Aldrich,>99%), V2O3 (Aldrich, 99.99%), and (NH4)H2PO4 (Alfa Aesar,>98%) were thoroughly mixed in a mortar with the aid ofisopropyl alcohol for 1 h. An excess of lithium (2%) was addedto compensate for high-temperature lithium evaporation. Afterthoroughly grinding the precursor materials, the dry mixturewas thermally decomposed at 450 °C in an Ar atmosphere for 1h. The decomposed and oxidized powder was once againground in a mortar and subsequently pressed into a pellet.Finally, the pellet was heated to 900 °C under a 95% Ar 5% H2gas flow for 5 h. Upon completion of the calcination process, agreen-colored pellet was obtained and ground into a finepowder for materials characterization.

XRD Characterization. Diffraction patterns were collectedon a Rigaku R-axis Spider diffractometer equipped with animage plate detector using a Cu Kα radiation source (λ = 1.542Å) operated at 40 kV and 40 mA. Samples were mounted on a0.5 mm Nylon loop and scanned for 10 min while rotating at1°/min. With the 2DP software, radial data were integratedover 2θ = 10−60°. The peaks were analyzed by using the JointCommittee on Powder Diffraction Standards (JCPDS) data-base.

Environmental Scanning Electron Microscopy (ESEM).Images were obtained with a model Quanta 650 FEG at 30.00kV.

Ex Situ and In Situ Raman Microscopy. Raman spectrawere acquired with a Renishaw inVia microscope system havinga 514.5 nm Ar+ laser in the backscattering configuration. For exsitu analysis, the beam was focused with a 50! objective lens,resulting in approximately a 1.3 μm spot diameter under air.The phase stability of Li3V2(PO4)3 under laser irradiation wasstudied at 0.067, 0.080, 0.113, 0.403, 0.855, 4.101, and 7.612mW power. The laser was focused on the same sample areawhile increasing or decreasing the power of the laser. For the insitu heat analysis, under N2 and air, a Linkam ScientificTHMS600 microscope stage was used to ramp the sampletemperature from room temperature to 600 °C. The temper-ature program used was as follows: 23−126 °C at a rate of 2°C/min, 126−150 °C at a rate of 1 °C/min, 150−195 °C at arate of 2 °C/min, 195−220 °C at a rate of 1 °C/min, and 220−600 °C at a rate of 10 °C/min. One hour temperature holdswere performed at 126, 150, 195, and 220 °C, whereas 30 mintemperature holds were performed at 258, 296, 334, 372, 410,448, 486, 524, 562, and 600 °C to ensure complete andhomogeneous heating of the sample. The beam was focusedwith an L 50! objective lens, resulting in approximately a 0.8μm spot diameter. An acquisition time of about 300 s was used

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200211995

for the in situ heat analysis and was varied between 20 and 120s for the ex situ studies.Thermal Studies. Differential scanning calorimetry (DSC)

was conducted on a TA Instruments DSC-Q100. A sample of9.000 mg was placed in an Al crucible. Each DSC cycleconsisted of heating the sample from 25 to 250 °C and thencooling it back to 25 °C. A total of three cycles were performedat a rate of 5 °C/min under a N2 flow rate of 50.00 mL/min.Thermal gravimetric analysis (TGA) was performed on a TAInstruments TGA-Q50. A sample of 4.347 mg was placed in aPt crucible. The first ramp consisted of heating the sample from25 to 500 °C at a rate of 2.50 °C/min. The second rampcontinued the heating from 500 to 800 °C at a rate of 5.00 °C/min.VASP-DFT. The vibrational frequency calculations were

performed with finite differences of the atomic forces fromDFT,19 using the HSE06 hybrid functional for the exchange-correlation energy.20 Valence electrons were described with aplane-wave basis set up to an energy cutoff of 283 eV; coreelectrons were incorporated into pseudopotentials in theprojector augmented wave phasework.21−24 Only the Γ pointwas included for the Brillouin-zone integration. The latticeparameters for the crystal supercell were set to experimentalvalues.2 Tests in which the volume was varied showed that theenergy of the experimental lattice constants was very close tothe minimum from fitting to the equation of state (0.15 eVhigher for 80 atoms). For the frequency calculations, onlysymmetrically inequivalent forward differences were taken tosave computational cost. The magnitude of each displacementwas 0.01 Å.The DFT+U functional was also performed with an effective

U value of 4.0 as a comparison, but the frequencies weresystematically underestimated in the whole range. Thefrequencies are eigenvalues of the Hessian matrix, whoseelements are the second derivatives of energy with respect tothe displacements. Thus, they are very sensitive to the accuracyof the potential energy surface. This is also the reason thatmakes Raman/IR spectroscopy a powerful tool in detectinglocal chemical environment changes.

3. RESULTS AND DISCUSSIONCharacterization of !-Li3V2(PO4)3. The solid-state syn-

thesis of α-Li3V2(PO4)3 was as described in full detail in the

Experimental and Theoretical Methods. The product wascharacterized with XRD and ESEM to determine the phasepurity, crystal structure, and particle size. As shown in Figure1a, the material can be indexed to the monoclinic space groupP21/n. ESEM images (Figure 1b) reveal highly agglomeratedparticles ranging from 0.1 to 50 μm in size.

Comparison of Experimental and Theoretical RamanSpectra. From the symmetry-based analysis of α-Li3V2(PO4)3,the predicted number of normal modes is 237 with 120 of thesenormal modes being Raman-active; the remainder are IR-active.7

Γ = + + +α‐ A B A B60 60 59 58g g u uLi V (PO )3 2 4 3 (1)

As with any centrosymmetric crystal, the IR active modes areRaman-inactive and vice versa. Any mode that respectscentrosymmetry is IR-active. The Raman-active modes can befurther labeled as Ag or Bg. Ag modes are invariant under all foursymmetry operations of the C2h point group, whereas Bg modesare invariant for two of the operations. By analyzing thesymmetry of each Raman mode for α-Li3V2(PO4)3 withpyspglib,25 all of the Raman modes were labeled. Some ofthe calculated Raman modes, along with their symmetryassignments, can be found in Table 1. A complete list of thecalculated modes can be found in Table S1 (SupportingInformation). The vibration modes corresponding to the 294.9,599.7, 1055.3, and 1085.7 cm−1 assignments are represented asschemes shown in Figure S1 (Supporting Information).26

From the unpolarized Raman spectrum of α-Li3V2(PO4)3,only 19 vibrational modes, 1077.9, 1058.7, 1030.0, 1009.2,975.4, 652.2, 603.6, 559.7, 505.5, 454.4, 430.4, 375.3, 349.4,290.4, 255.5, 223.9, 169.4, 134.1, and 118.1 cm−1, weredetected (Figure 2 and Table 1).Nevertheless, the DFT calculations clearly distinguish

between the experimental spectral regions where vibrationalmodes occur from the regions where no vibrations are present.Vibrational spectra of transition-metal phosphates consist ofinternal and external modes. Internal modes are vibrations thatoccur within atomic groups (such as LiO4, MO6, and PO4polyhedra), and external modes are vibrations of the atomicgroups and crystal lattice.17 For the Li3M2(PO4)3 compounds,the PO4 valency bond vibrations occur between 1250 and 900cm−1, PO4 deformation vibrations are in the 700−400 cm−1

region, and the external lattice vibrations correspond to the

Figure 1. (a) XRD and (b) ESEM of microcrystalline Li3V2(PO4)3.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200211996

low-frequency region of 400−0 cm−1. As expected, the PO4valency bond vibrations are the most intense peaks.Vibrational Modes of !-Li3V2(PO4)3 as a Function of

Laser Irradiation Power. Raman spectra of α-Li3V2(PO4)3were taken at different laser irradiation power to examine thephase stability of the material. At 0.067 mW, peaks centered at1058.7, 1030.0, 1009.2, 975.4, 603.6, 454.4, 430.4, 290.4, 223.9,169.4, and 134.1 cm−1 were detected, as is shown in Figure 3a.At 0.113 mW, additional peaks at 1077.9, 652.2, 559.7, 505.5,375.3, 349.4, 255.5, and 118.1 cm−1 were detected. A gradualincrease in power to 0.855 mW resulted in these peaksbecoming sharper and more intense with no additional featuresappearing in any of the spectra (Figure 3a). Therefore, it was

concluded that, within the 0.067−0.855 mW power range, themonoclinic phase was conserved.Irradiation at 0.855−4.101 mW resulted in various changes

of the Raman spectrum (Figure 3b,c). Because spectral changesoccurred rapidly, the acquisition time was reduced 6-fold in thispower range. Within the first scan (20 s), the originally sharpphosphate vibrations in the high region of the spectrum werereplaced with two broad peaks at 1048 and 1001 cm−1. Twooriginally distinct peaks at 454.4 and 430.3 cm−1 merged into asingle peak. Additionally, a shoulder around 1130 cm−1

emerged and the lattice vibrations significantly diminished inintensity. By the fifth scan (Figure 3b), the shoulder at 1130cm−1 continued to grow in intensity and experienced a slightblue shift. Two new peaks were found at 947 and 876 cm−1.After 220 s (11 scans), the new features of the spectrumstabilized (the shoulder at 1130 cm−1 was now further blueshifted to 1140 cm−1). To determine whether the features ofthe stabilized spectrum were irreversible, the power wasreduced to 0.855 mW. After seven scans, none of the generalfeatures changed (Figure S2a, Supporting Information).Therefore, only one scan was conducted at the lower irradiationpower of 0.403 and 0.113 mW (Figure S2b, SupportingInformation). Clearly, the original spectrum was not recovered.Continuous irradiation at 4.101 mW caused the 876 cm−1

mode to dramatically increase in intensity (Figure 3d, i and ii).At the highest irradiation power of 7.612 mW, the intensity ofthe 876 cm−1 peak dominated the remaining spectral features ofthe spectrum (Figure 3d). Moreover, visually the original greenpowder became a much darker, olive green. Evidently, α-Li3V2(PO4)3 is not stable at an irradiation power of 4.101 mWor greater.Similar power studies have been conducted on olivine

LiFePO4. Bai et al.27 showed that the thermal effect of laser

irradiation created two new peaks in the low-wavenumberregion at 215 and 277 cm−1. These peaks were associated withformation of α-Fe2O3.

28 Continual laser irradiation led to abroad background in the high-wavenumber region that theauthors attributed to amorphous FePO4. However, uponreversing the power of the laser, the amorphous backgroundvanished and the most prominent peak centered at 950 cm−1

remained unchanged in intensity, showing that the basicstructure was maintained throughout the course of the study. Itcould be assumed that comparable transformations occurred forα-Li3V2(PO4)3, possibly leading to the creation of V2O5 andamorphous V2(PO4)3. The formation of these phase impuritiescreated by laser irradiation on Li3V2(PO4)3 were indirectlyinvestigated through thermal studies.

Thermal Studies on !-Li3V2(PO4)3. In the confocal Ramanmicroscope arrangement, a very high power density isconcentrated on a small area of the sample. Absorption ofradiation can cause the local temperature to increase by severalhundreds of degrees. Consequently, as the irradiation powerwas increased, the local temperature of α-Li3V2(PO4)3 alsoincreased. Having conducted the measurements in air, it isreasonable to suspect that the material is susceptible tooxidation at the higher irradiation power and that oxidationbrought about the changes in the Raman spectrum. In situRaman heat treatment in air and in an inert atmosphere (N2)clarified this using a temperature-controlled heating stage.A microscope heating stage permitted the acquisition of

Raman spectra for α-Li3V2(PO4)3 from room temperature to600 °C. No spectral changes were observed from 23.5 to 116.7°C (Figure 4a), indicating that the monoclinic phase was

Table 1. Raman Modes for Li3V2(PO4)3 Observed by RamanMicroscopy and the Corresponding Calculated RamanModes with Symmetry Assignments by VASP-DFT

Raman vibrational modes for Li3V2(PO4)3 in units of cm−1

number calculated observed assignment

1 1085.7 1077.9 Ag

2 1055.3 1058.7 Bg

3 1030.4 1030.0 Ag

4 1010.2 1009.2 Ag

5 974.0 975.4 Ag

6 654.9 652.2 Ag

7 599.7 603.6 Bg

8 560.4 559.7 Ag

9 508.0 505.5 Bg

10 451.2 454.4 Bg

11 427.5 430.4 Bg

12 376.5 375.3 Ag

13 346.1 349.4 Ag

14 294.9 290.4 Ag

15 252.7 255.5 Ag

16 224.3 223.9 Ag

17 167.6 169.4 Ag

18 136.2 134.1 Ag

19 116.7 118.1 Bg

Figure 2. Observed (top) and calculated (bottom) Raman spectra forα-Li3V2(PO4)3 in the 50−1230 cm−1 wavenumber range. Theobserved spectrum above was acquired with a 50! objective at apower of 0.855 mW and a total acquisition time of 120 s at roomtemperature.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200211997

preserved in air within this temperature range. At 125 °C, the1030 cm−1 peak and shoulder at 1079 cm−1 disappeared. Thesegeneral features remained until reaching 202.0 °C, at whichpoint more dramatic changes took place, such as broadening ofthe 1009 cm−1 peak, merging of the 435 and 455 cm−1 peaksinto a single peak, transformation of the 979 cm−1 peak into ashoulder, and a significant diminishing of the lattice modes. Thegeneral features of the spectrum at 202.0 °C (Figure 4b) closelymatched those of the first scan at a power of 4.101 mW in theirradiation power study (Figure 3b).

A further increase in heating temperature (Figure 4c)resulted in additional spectral changes. At 333.7 °C, anadditional peak at 1140 cm−1 can be seen that gradually grewin intensity with higher heating temperatures and experienced aslight blue shift once reaching 595.8 °C. Furthermore, at 595.8°C, a new peak at 876 cm−1 became the dominant mode of thespectrum. These changes are analogous to the spectraltransitions seen during the irradiation power study (FigureS3a,b, Supporting Information). However, the lower intensityof the 876 cm−1 mode at 595.8 °C as compared to the spectrumat 7.612 mW indicated two things: the local temperature of the

Figure 3. Raman spectra of α-Li3V2(PO4)3 at different laser irradiation power with the 50! objective: (a) from 0.067 to 0.855 mW (120 s acquisitiontime), (b, c) 11 successive measurements at 4.101 mW, with each acquisition being 20 s, (d) continuous exposure at 4.101 mW with longeracquisitions of (i) 120 s and (ii) 300 s and the highest power measurement at 7.612 mW (120 s acquisition time).

Figure 4. In situ Raman spectra of α-Li3V2(PO4)3 at different heating temperatures under air: (a) from 23.5 to 116.7 °C, (b) 125.0 to 276.4 °C, and(c) 333.7 to 595.8 °C taken with a 50! objective and a laser power of 0.855 mW for 300 s.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200211998

sample under 7.612 mW power was greater than 600 °C, and at600 °C, the material was not completely oxidized.The in situ Raman heating experiment was repeated under a

N2 atmosphere. This experiment was done to confirm that α-Li3V2(PO4)3 underwent oxidation under either elevatedtemperatures or irradiation power in air and to distinguishany spectral, and therefore structural changes, that could beunrelated to the oxidation reaction. Such structural changesinclude transformation of lattice symmetry, cell expansion/contraction, and atomic disordering. Under N2, the monoclinicphase is also retained within the 22.3−108.2 °C temperaturerange (Figure 5a). Again, no appreciable peak broadening isnoted. As in the heating experiment conducted in air, at 126.1°C, the 1030 cm−1 peak and shoulder at 1079 cm−1

disappeared. The same spectral characteristics detected at202.0 °C in air were detected at 195.0 °C in N2, which includedbroadening of the 1009 cm−1 peak, merging of the 435 and 455cm−1 modes into a single peak, transformation of the 979 cm−1

peak into a shoulder, and lowered intensity of the lattice modes(Figure 5b). Increasing the temperature from 195.0 to 600.0 °Cresulted in no further changes in the spectrum (Figure 5c).When the material was subsequently cooled down to roomtemperature, the Raman spectrum exactly matched that of theinitial α-Li3V2(PO4)3 (Figure S4, Supporting Information).Li3V2(PO4)3 is isotypic with Li3M2(PO4)3, where M = Fe, Sc.

These crystals are found in the α-phase at room temperatureand exhibit two structural phase changes at elevated temper-atures (β and γ).29−31 In the α ! β transition, the MO6octahedra and PO4 tetrahedra show a slightly higher symmetry,but the greatest distinction in the structure is the Li ionarrangement.29 In the α-phase, Li ions are located in threeunique positions, Li(1), Li(2), and Li(3). The least energeti-cally favorable, Li(3), is repositioned to a new location, Li(1s),which is related in symmetry to Li(1). During the high-temperature transition from β ! γ, more notable structuraltransformations occur. The MO6 octahedra and PO4 tetrahedra3-D framework converts from the monoclinic to theorthorhombic structure, and all of the Li ions are rearranged.The 12 Li ions of the unit cell are repositioned over three newcrystallographic positions where two of the positions are onlypartially occupied. It is expected that α-Li3V2(PO4)3 alsoundergoes similar intermediate and high-temperature phase

transitions. Differential scanning calorimetry (DSC) studies ofα-Li3V2(PO4)3 were completed from 25 to 250 °C and thenback to 25 °C for a total of three cycles. The first cycle was usedto remove any adsorbed water from the material, and the nexttwo cycles were completed to determine the reversibility of thephases. As shown in Figure 6, two structural modifications

occur at roughly 123 °C (β-phase) and 191 °C (γ-phase).28,29

All of the phase transitions were reversible upon cooling backto room temperature (Figure 6, and Figure S5a,b, SupportingInformation).From the DSC plot, the temperature for the crystalline phase

changes of Li3V2(PO4)3 occurred at 123 °C (α ! β) and 191°C (β ! γ), and from 191 °C and greater, the γ-phase ispredominant. Undeniably, these crystalline phase changes areresponsible for the Raman spectral changes between 23 and200 °C under N2 and air (Figure 7). However, in situ XRD dataprovided by Bykov et al.29 have shown evidence of high-temperature twinning of α + β, β + γ, and even of all threephases at once. Therefore, the spectra in Figure 7 may notexclusively represent a single phase. In the β ! γ transition,there is a more drastic change in structure as compared to the α! β transition. From the Raman spectral changes of the phases,the opposite would be expected. Twinning of the phases couldexplain why the structural and Raman trends do not follow.

Figure 5. In situ Raman spectra of α-Li3V2(PO4)3 at different temperatures under N2: (a) from 22.3 to 108.2 °C, (b) 126.1 to 220.0 °C, and (c)334.6 to 600.0 °C taken with a 50! objective and a laser power of 0.855 mW for 300 s.

Figure 6. DSC curve of α-Li3V2(PO4)3 from 25 to 250 °C at a rate of5 °C/min under a N2 flow rate of 50.00 mL/min.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200211999

No crystallographic phase transitions for Li3V2(PO4)3occurred past 200 °C. Therefore, any Raman spectral changespast this temperature in air must have been induced byoxidation. To prove this hypothesis, TGA under an air flow wasused to monitor the weight change of Li3V2(PO4)3 from 25 to800 °C (Figure 8a). The original 4.437 mg sample initially lost0.024 mg, corresponding to the evaporation of adsorbed water.After 330 °C, the material began to oxidize, and by 800 °C, ithad an ∼4% increase in mass.32 In this case, the originally loose,green powder had sintered into a single, large, olive-coloredparticle.XRD was conducted thereafter to determine phase

impurities. No V2O5 was detected. Instead, the material wasfound to be triclinic LiVOPO4 (P1 space group) with noLi3V2(PO4)3 detected from the pattern (Figure 8b). TheRaman spectrum included vibrational modes at 1116, 1039,

1028, 993, 956, 685, 615, 536, 332, 255, 148, and 119 cm−1

(Figure 8c). To the best of our knowledge, no Raman literatureexists for triclinic α-LiVOPO4. Raman data are, however,reported for rhombohedral β-LiVOPO4.

33 Qualitatively speak-ing, the spectrum for β-LiVOPO4 correlates well with theRaman spectrum of the oxidized Li3V2(PO4)3 material. The twospectra share the dominating spectral feature as an intense peakclose to ∼880 cm−1 that Baran et al.33 assigned to the VOstretching mode. However, the possibility of low concentrationsfor the partially delithiated phases Li3−xV2(PO4)3, 0 < x < 3,possibly undetectable by XRD, cannot be ruled out.Interestingly, Kuo et al.34 found through in situ XRD studiesthat α-LiVOPO4 transforms into α-Li3V2(PO4)3 at temper-atures above 600 °C under a reducing atmosphere. The presentwork has identified a new reaction. α-Li3V2(PO4)3 transformsinto α-LiVOPO4 at temperatures above 600 °C in an oxidizingatmosphere.Understanding the behavior of a species under laser

irradiation and heating is critical for correct interpretation ofmicro-Raman experiments. Previously, Burba et al.7 conductedRaman vibrational studies of α-Li3V2(PO4)3 in the hopes ofconducting in situ spectroelectrochemical measurements of thismaterial. This study was particularly interested in following thehigh-frequency region of the spectrum assigned to PO4 valencystretching modes, which are the most intense peaks in thespectrum. However, there were only four weak bands detectedin the 1200−850 cm−1 region at 1142, 1055, 1011, and 973cm−1. The spectrum of α-Li3V2(PO4)3 at 411.1 °C also showssimilarly broad peaks at 1138, 1053, 1005, and 977 cm−1

(Figure 5c). It is reasonable to suspect that the features foundby Burba et al.7 correspond to thermal degradation of the α-phase into the γ-phase, with the peak at 1138 cm−1 indicatingoxidation.

Figure 7. Raman spectra of the α, β, and γ crystalline phases for α-Li3V2(PO4)3 under N2.

Figure 8. (a) TGA curve of α-Li3V2(PO4)3 under air from 25 to 800 °C. The first ramp was from 25 to 500 °C at a rate of 2.50 °C/min. The secondramp continued heating from 500 to 800 °C at a rate of 5.00 °C/min. (b) XRD and (c) Raman of the sample after the TGA analysis with the 50!objective at a power of 0.855 mW with a 120 s acquisition time identifying α-LiVOPO4 as the product of oxidation.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200212000

4. CONCLUSIONSIn this paper, we report the experimental Raman spectrum,along with the corresponding theoretical Raman spectrum andsymmetry assignments for α-Li3V2(PO4)3 by DFT. Exper-imentally, only 19 of the 120 Raman modes predicted by factorgroup analysis were identified. As anticipated, the internalphosphate stretching modes located in the high-frequencyregion contain the sharpest and most intense peaks, making itthe prevailing feature of the spectrum. Like LiFePO4,Li3V2(PO4)3 is unstable under certain laser irradiation power(∼≥4 mW) caused by rapid heating and a temperature increaseunder micro-Raman measurements. From our Raman andthermal measurements, it was found that the monoclinic phaseexperiences two thermally induced structural transformations tothe β- and γ-phases analogous to its isostructural counterparts,Li3Sc2(PO4)3 and Li3Fe2(PO4)3. Continuous heating under airresults in oxidation of α-Li3V2(PO4)3 to α-LiVOPO4. Theseexperiments provide fundamental groundwork for future in situRaman studies of α-Li3V2(PO4)3 as a cathode for lithium-ionstorage devices.

! ASSOCIATED CONTENT*S Supporting InformationA table and five figures are provided as mentioned in the text.This material is available free of charge via the Internet athttp://pubs.acs.org.

! AUTHOR INFORMATIONCorresponding Author*Phone: + 1-512-232-9160 (K.J.S.), + 1-512-471-4179 (G.H.).Fax: + 1-512-471-8696 (K.J.S.), + 1-512-471-8696 (G.H.). E-mail: stevenson@cm.utexas.edu (K.J.S.), henkelman@mail.utexas.edu (G.H.).NotesThe authors declare no competing financial interest.

! ACKNOWLEDGMENTSThis material is based upon work supported as part of theprogram “Understanding Charge Separation and Transfer atInterfaces in Energy Materials (EFRC:CST)”, an EnergyFrontier Research Center funded by the U.S. Department ofEnergy, Office of Science, Office of Basic Energy Sciences,under Award Number DE-SC0001091. We also thank theWelch Foundation (Grant F-1529) for financial support. Wealso would like to thank Donald A. Robinson for acquiring theESEM image.

! REFERENCES(1) Yuan, L.-X.; Wang, Z.-H.; Zhang, W.-X.; Hu, X.-L.; Chen, J.-T.;Huang, Y.-H.; Goodenough, J. B. Development and Challenges ofLiFePO4 Cathode Material for Lithium-Ion Batteries. Energy Environ.Sci. 2011, 4, 269−284.(2) Yin, S.-C.; Grondey, H.; Strobel, P.; Anne, M.; Nazar, L. F.Electrochemical Property: Structure Relationships in MonoclinicLi3V2(PO4)3. J. Am. Chem. Soc. 2003, 125, 10402−10411.(3) Huang, H.; Yin, S.-C.; Kerr, T.; Taylor, N.; Nazar, L. F.Nanostructured Composites: A High Capacity, Fast Rate Li3V2-(PO4)3/Carbon Cathode for Rechargeable Lithium Batteries. Adv.Mater. 2002, 14, 1525−1528.(4) Lee, S.; Park, S. S. Atomistic Simulation Study of MonoclinicLi3V2(PO4)3 as a Cathode Material for Lithium Ion Battery: Structure,Defect Chemistry, Lithium Ion Transport Pathway, and Dynamics. J.Phys. Chem. C 2012, 116, 25190−25197.

(5) Baddour-Hadjean, R.; Pereira-Ramos, J.-P. Raman Micro-spectrometry Applied to the Study of Electrode Materials for LithiumBatteries. Chem. Rev. 2009, 110, 1278−1319.(6) Morcrette, M.; Leriche, J.-B.; Patoux, S.; Wurm, C.; Masquelier,C. In Situ X-Ray Diffraction During Lithium Extraction fromRhombohedral and Monoclinic Li3V2(PO4)3. Electrochem. Solid-StateLett. 2003, 6, A80−A84.(7) Burba, C. M.; Frech, R. Vibrational Spectroscopic Studies ofMonoclinic and Rhombohedral Li3V2(PO4)3. Solid State Ionics 2007,177, 3445−3454.(8) Shu, J.; Shui, M.; Xu, D.; Gao, S.; Yi, T.; Wang, D. J.; Li, X.; Ren,Y. Design and Comparison of Ex Situ and In Situ Devices for RamanCharacterization of Lithium Titanate Anode Material. Ionics 2011, 17,503−509.(9) Burba, C. M.; Frech, R. Modified Coin Cells for In Situ RamanSpectroelectrochemical Measurements of LixV2O5 for LithiumRechargeable Batteries. Appl. Spectrosc. 2006, 60, 490−493.(10) Kim, Y. A.; Kojima, M.; Muramatsu, H.; Umemoto, S.;Watanabe, T.; Yoshida, K.; Sato, K.; Ikeda, T.; Hayashi, T.; Endo, M.;Terrones, M.; Dresselhaus, M. S. In Situ Raman Study on Single- andDouble-Walled Carbon Nanotubes as a Function of Lithium Insertion.Small 2006, 2, 667−676.(11) Cahill, L. S.; Chapman, R. P.; Britten, J. F.; Goward, G. R. 7LiNMR and Two-Dimensional Exchange Study of Lithium Dynamics inMonoclinic Li3V2(PO4)3. J. Phys. Chem. B 2006, 110, 7171−7177.(12) Yin, S.-C.; Grondey, H.; Strobel, P.; Huang, H.; Nazar, L. F.Charge Ordering in Lithium Vanadium Phosphates: ElectrodeMaterials for Lithium-Ion Batteries. J. Am. Chem. Soc. 2002, 125,326−327.(13) Patoux, S.; Wurm, C.; Morcrette, M.; Rousse, G.; Masquelier, C.A Comparative Structural and Electrochemical Study of MonoclinicLi3Fe2(PO4)3 and Li3V2(PO4)3. J. Power Sources 2003, 119−121, 278−284.(14) Burba, C. M.; Frech, R. Raman and FTIR Spectroscopic Studyof LixFePO4 (0 ≤ x ≤ 1). J. Electrochem. Soc. 2004, 151, A1032−A1038.(15) Barj, M.; Lucazeau, G.; Delmas, C. Raman and Infrared Spectraof Some Chromium Nasicon-Type Materials: Short-Range DisorderCharacterization. J. Solid State Chem. 1992, 100, 141−150.(16) Butt, G.; Sammes, N.; Tompsett, G.; Smirnova, A.; Yamamoto,O. Raman Spectroscopy of Superionic Ti-Doped Li3Fe2(PO4)3 andLiNiPO4 Structures. J. Power Sources 2004, 134, 72−79.(17) Kravchenko, V. V.; Mikhailov, V. I.; Sigarev, S. E. Some Featuresof Vibrational Spectra of Li3M2(PO4)3 (M = Sc, Fe) Compounds Neara Superionic Phase Transiton. Solid State Ionics 1992, 50, 19−30.(18) Paraguassu, W.; Freire, P. T. C.; Lemos, V.; Lala, S. M.;Montoro, L. A.; Rosolen, J. M. Phonon Calculation on Olivine-likeLiMPO4 (M = Ni, Co, Fe) and Raman Scattering of the Iron-Containing Compound. J. Raman Spectrosc. 2005, 36, 213−220.(19) Kohn, W.; Becke, A. D.; Parr, R. G. Density Functional Theoryof Electronic Structure. J. Phys. Chem. 1996, 100, 12974−12980.(20) Heyd, J.; Scuseria, G. E.; Ernzerhof, M. Hybrid FunctionalsBased on a Screened Coulomb Potential. J. Chem. Phys. 2003, 118,8207−8215.(21) Kresse, G.; Furthmuller, J. Efficiency of Ab-Initio Total EnergyCalculations for Metals and Semiconductors Using a Plane-Wave BasisSet. Comput. Mater. Sci. 1996, 6, 15−50.(22) Kresse, G.; Furthmuller, J. Efficient Iterative Schemes for AbInitio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys.Rev. B 1996, 54, 11169−11186.(23) Blochl, P. E. Projector Augmented-Wave Method. Phys. Rev. B1994, 50, 17953−17979.(24) Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to theProjector Augmented-Wave Method. Phys. Rev. B 1999, 59, 1758−1775.(25) pyspglib for ASE; 2013. http://spglib.sourceforge.net/pyspglibForASE/ (Accessed February 15, 2013).

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200212001

(26) Canepa, P.; Hanson, R. M.; Ugliengo, P.; Alfredsson, M. J-ICE:A New Jmol Interface for Handling and Visualizing Crystallographicand Electronic Properties. J. Appl. Crystallogr. 2011, 44, 225−229.(27) Bai, Y.; Yin, Y.; Yang, J.; Qing, C.; Zhang, W. Raman Study ofPure, C-Coated and Co-Doped LiFePO4: Thermal Effect and PhaseStability Upon Laser Heating. J. Raman Spectrosc. 2011, 42, 831−838.(28) Shebanova, O. N.; Lazor, P. Raman Study of Magnetite(Fe3O4): Laser-Induced Thermal Effects and Oxidation. J. RamanSpectrosc. 2003, 34, 845−852.(29) Bykov, A. B.; Chirkin, A. P.; Demyanets, L. N.; Doronin, S. N.;Genkina, E. A.; Ivanov-Shits, A. K.; Kondratyuk, I. P.; Maksimov, B. A.;Mel’nikov, O. K.; Muradyan, L. N.; Simonov, V. I.; Timofeeva, V. A.Superionic Conductors Lithium Metal Phosphate (Li3M2(PO4)3)(M =Fe, Sc, Cr): Synthesis, Structure and Electrophysical Properties. SolidState Ionics 1990, 38, 31−52.(30) Ohkawa, H.; Yoshida, K.; Saito, M.; Uematsu, K.; Toda, K.;Sato, M. Improvement of Discharge Capacity of β-Fe2(SO4)3-TypeLi3V2(PO4)3 by Stabilizing High Temperature Orthorhombic Phase atRoom Temperature. Chem. Lett. 1999, 28, 1017−1018.(31) Sato, M.; Ohkawa, H.; Yoshida, K.; Saito, M.; Uematsu, K.;Toda, K. Enhancement of Discharge Capacity of Li3V2(PO4)3 byStabilizing the Orthorhombic Phase at Room Temperature. Solid StateIonics 2000, 135, 137−142.(32) Kuang, Q.; Zhao, Y.; An, X.; Liu, J.; Dong, Y.; Chen, L.Synthesis and Electrochemical Properties of Co-Doped Li3V2(PO4)3Cathode Materials for Lithium-Ion Batteries. Electrochim. Acta 2010,55, 1575−1581.(33) Baran, E. J.; Vassallo, M. B.; Lii, K. H. Vibrational Spectra of β-LiVOPO4 and NaVOPO4. J. Raman Spectrosc. 1994, 25, 199−202.(34) Kuo, H. T.; Bagkar, N. C.; Liu, R. S.; Shen, C. H.; Shy, D. S.;Xing, X. K.; Lee, J.-F.; Chen, J. M. Structural Transformation ofLiVOPO4 to Li3V2(PO4)3 with Enhanced Capacity. J. Phys. Chem. B2008, 112, 11250−11257.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp403282a | J. Phys. Chem. C 2013, 117, 11994−1200212002